Google unveils Gemini Canvas: multimodal model for interactive prototyping and asset generation

AI · 5 min read

Gemini Canvas supports layered outputs: designers can prompt with a sketch and text instructions, and the model returns an editable prototype with components wired to simple interactions. It also generates asset variants and recommended accessibility improvements.

Collaboration features include live suggestion tiles that teammates can accept into the prototype and an audit trail that explains how decisions were derived from prompt context. Google highlighted on-device privacy options for enterprise customers.

The company is integrating Gemini Canvas into Workspace and offering plugins for popular design tools. Pricing and enterprise controls were announced for later this quarter, with beta access for selected partners.